Should I Believe It?
3
교육 | 2.6MB
기술
The "Should I Believe It?" (¿Me Lo Creo?) app lets you see how often you will be right or wrong if you believe the positive results of randomized controlled trial or other quantitative study, given a particular p-value threshold and prior probability that the treatment will work or the hypothesis be true.
"Should I Believe It?" helps users understand or teach statistical concepts, in particular the p-value. It shows how the power of a study and the prior probability of a hypothesis being true influence the interpretation of a study's findings.
The app incorporates Bayesian reasoning and is useful for explaining statistics to people. It can, for example, show why the results of an unbiased trial in which a treatment is associated with a clinically important outcome at a low p-value (that is which is also statistically significant) may not be convincing evidence of effective if the prior probability of effectiveness is low (such as homoeopathic treatments where the degree of dilution means there are probably no orginal molecules still present).
Show More
Less